On this page
Five numbers: visitors, where they came from, how many joined your list, how many bought, and which single page did best. Ten minutes, once a week, written down. Everything else a dashboard offers is either downstream of those five or too noisy to read at the size most businesses actually are.
The point of the list is not completeness. It is that each of the five changes a decision, and the ones left out do not.
Why five, and why weekly?
Because on a small site, daily numbers are mostly weather.
One person sharing a link in a group chat moves a daily figure more than a week of your work does. Look at it every morning and you will spend the month reacting to strangers. Look at it monthly and you will have published eight more things before you learn whether the first one landed.
A week averages out the accident and is still short enough to change what you do next.
The five
| # | Number | The decision it changes |
|---|---|---|
| 1 | Visitors | whether to keep publishing or start distributing |
| 2 | Where they came from | where to spend next week's effort |
| 3 | New email subscribers | whether the free thing is worth its page |
| 4 | Sales | whether anything is actually working |
| 5 | Best single page | what to write more of |
Visitors, not sessions or pageviews. One number, same definition every week.
Where they came from is the most useful of the five and the one most people glance past. A referrer that repeats is a channel; a referrer that appears once is an accident. At the start you are looking for the first repeat, not for volume.
New subscribers tells you whether the free thing is doing its job. Visitors without signups usually means the offer is not visible rather than not wanted — check whether the form is on the pages people actually land on.
Sales. Zero is a number. It is early, it is common, and it stops being acceptable as an answer once the first four numbers are healthy.
Best single page is the one that tells you what to make next. Not your favourite page — the one that got read.
The Short Version
Every conclusion from all eight playbooks — 97 chapters, condensed. Free.
What should you leave out?
Everything that goes up on its own.
| Number | Why it misleads |
|---|---|
| Total pageviews since launch | only ever increases; cannot go down, so cannot inform |
| Followers | the account can grow while nobody clicks |
| Impressions | measures what a platform showed, not what anyone did |
| Bounce rate on articles | reading the answer and leaving is a success |
| Time on page | mostly measures tab-switching |
The test for any metric: if this number went up thirty per cent, would I do something differently? If the answer is no, it is not a number — it is a mood.
Bounce rate deserves its own note, because it is the most commonly misread figure in this category. On a checkout it is meaningful. On an article that answers a question well, a high bounce rate is what success looks like: they came, they got it, they left.
What if every number is zero?
Then four of the five are not the useful ones this week, and the second is.
At the beginning the question is not "how many" but "from where". Ten visitors from one place that keeps sending them is a channel you can grow. Two hundred from a single link that will never repeat is a nice afternoon.
So the early version of this review is smaller: where did the handful come from, and did any source appear twice? That is it. Chasing the volume number before you have an answer to that is how people end up publishing for a year into nothing.
How do you actually do the check?
Write it down. Not a screenshot — a line of five numbers with the date.
A dashboard shows you now. What you need is the comparison against last week, and your memory of last week is worse than you think. A plain file with one row per week does more for this than any tool, and it keeps working when you change analytics providers.
| Week | Visitors | Top source | Subs | Sales | Best page |
|---|---|---|---|---|---|
| 2026-W41 |
Ten minutes. If it takes longer, you have started reading instead of deciding.
What the numbers will not tell you
Which of the five is the bottleneck is usually obvious. Why it is the bottleneck is not.
Visitors but no subscribers, subscribers but no sales, sales but no repeat — each of those is a different problem and the dashboard says nothing about which. That answer comes from asking people, and from did your content do anything?, which is about tracing an effect back to a cause when you have no budget for proper attribution.
Picking the tool to get these five out of comes first: Plausible, Umami or Vercel Analytics covers which free tier can actually see a signup.
The AI Mastery Playbook is the one that covers building the weekly habit rather than the dashboard.
Tools in this piece
| Tool | What it does | Price |
|---|---|---|
| Plausible | Simple, privacy-friendly visitor stats with no cookie banner. | ~€9/mo |
| Umami | Same idea, free if you host it yourself. | Free self-hosted |
| Vercel | Hosting. Puts a project live on a real address with one command. | Free tier |
| Stripe | Payments and subscriptions. The standard, and integrating it is a matter of asking. | Free, percentage per sale |
| Kit | Email platform for creators. The list is the only thing you actually own. | Free under 1,000 subs |
Every tool has its own page with the price, who should skip it and what to check before paying. Some links are affiliate links.
Questions people ask
Which metrics should a small business check weekly?
Five: visitors, where they came from, how many joined your list, how many bought, and which single page did best. Everything else is either downstream of those or too noisy to read week to week.
Why is weekly the right interval?
Daily numbers on a small site are mostly noise — one person sharing a link moves them more than anything you did. Monthly is too slow to change what you publish next. A week is long enough to average out an accident and short enough to act on.
Is bounce rate worth tracking?
Rarely, on a content site. Someone who reads an article, gets the answer and leaves is a success that bounce rate records as a failure. It is a useful number on a checkout and a misleading one on a blog post.
What is a vanity metric?
Any number that goes up when nothing improves. Followers, impressions and total pageviews since launch all qualify. The test is simple: if the number went up 30%, would you do anything differently? If not, stop looking at it.
What should you do when every number is zero?
Check the only one that is not — where the few visitors came from. At the start, direction matters more than volume, and the first referrer that repeats is more useful than a hundred that do not.
How long should the weekly check take?
Ten minutes. If it takes longer you are reading, not deciding. Write the five numbers down somewhere permanent so the comparison is against last week rather than against your memory of last week.